Video Coding History-Based Motion Vector Predictor Candidate List Management
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing picture coding and decoding technologies face high processing loads due to the complexity of affine transforms used in inter prediction, which hinders efficient coding efficiency.
Innovation Solution
A picture coding device and method that includes a coding information storage unit, a merging candidate list generation unit, a history-based merging candidate addition unit, and a merging candidate selection unit to efficiently manage motion vector predictor candidates and reduce processing load by filling history-based candidate lists with predetermined motion vectors and reference indices, and selecting optimal merging candidates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If affine transform is applied in inter prediction, then coding efficiency is improved, but processing load increases
Solution Approach 1:
The patent divides the picture into multiple blocks and performs prediction processing in units of blocks. By segmenting the processing into smaller block-level operations with history-based candidate lists, the complex affine transform is broken down into manageable segments that reduce overall processing load while maintaining coding efficiency.
Solution Approach 2:
The patent fills history-based motion vector predictor candidate lists with predetermined motion vectors and reference indices before the actual prediction process. This preliminary preparation of candidate lists reduces the computational burden during real-time encoding by having ready-made candidates available for selection, thereby lowering processing load while preserving coding performance.
2Device complexity
If history-based candidate lists are filled with predetermined motion vectors, then processing load is reduced, but coding efficiency may be compromised
Solution Approach 1:
The patent implements a feedback mechanism where the history-based motion vector predictor candidate lists are dynamically updated and refined based on previous prediction results. The candidate lists are filled with predetermined vectors that are continuously improved through feedback from actual encoding performance, ensuring that processing load is reduced without sacrificing coding efficiency.
Solution Approach 2:
The patent changes the parameters of motion vector prediction by using history-based candidate lists with predetermined motion vectors and reference indices. This parameter change allows the system to operate with simplified prediction models that reduce processing load while maintaining accurate motion compensation through optimized candidate selection.
Data Source
AI summary
A picture coding device includes a coding information storage unit configured to initialize to fill a history-based motion vector predictor candidate list with history-based candidates having at least predetermined motion vector and reference index, a motion vector predictor candidate derivation unit configured to derive a history-based candidate included in the history-based motion vector predictor candidate list as a motion vector predictor, a motion vector subtraction unit configured to calculate a motion vector difference by subtracting the derived motion vector predictor from a motion vector by which inter prediction is performed, wherein the coding information storage unit adds the motion vector to the history-based motion vector predictor candidate list as a history-based candidate and updates the history-based motion vector predictor candidate list.